Machine learning is an algorithm that allows computers to learn from existing data to find patterns and apply the learned results to new data to predict the future. Practical examples include machine translation, speech recognition, dialogue systems, handwriting recognition, facial image recognition, and automated driving. Recent research and applications in the medical field include automated detection of tumors on CT images and endoscopic images.

　The advantages of machine learning are that it can be applied to a wide range of problems and that large amounts of data that can ordinarily not be handled by humans can be processed, as it is more accurate and faster than humans.

　A practical example of machine learning in the field of equilibrium research is the prediction of vestibular disorders from posturography data. We evaluated a dataset from Fujimoto et al. (Otol. Neurotol., 2014), including posturography and vestibular function data and found that machine learning algorithms can be successfully used to predict vestibular dysfunction as identified using caloric testing with the dataset of the center of pressure sway during posturography.

　Some of the points to be considered for practical application of machine learning in the field of vertigo research include the following: clinical data contain many errors, and database errors may occur frequently, the accuracy of clinical examinations should be taken into account, the difference between the acute and chronic phases of disease should be taken into account, and the dizziness symptom varies among cases. In order to achieve better accuracy, a large amount of data is required, and multi-institutional joint research should be considered.